The Museum of Human Language

Several years ago, a TV commercial claimed that “without chemistry, life itself would be impossible.”  In reality, the commercial should have said “without chemicals”, not “without chemistry.”    Life is made of chemicals, not the science of chemicals.  So what is a science, such as chemistry?   Science consists of the activities of scientists.   The knowledge, facts and theories of science would be meaningless without scientists to interpret them.   Of course, written language is the most important means by which scientists share their ideas.   But books expounding scientific theories are useless unless they are interpreted by people, just like a lecture in astrophysics given to a group of preschoolers. Mathematics, too, is an important means by which scientists communicate their ideas, but unless the mathematics is understood, the relation of the mathematics and the data is pointless. 

But there is an important difference between mathematics and language -- mathematicians are allowed and encouraged to invent their own systems, but linguists and other language users hardly ever are lucky enough even to coin their own phrases, much less create major grammatical systems.   There are some parallels between mathematical and linguistic innovation, but they are fundamentally different processes.   A mathematical paper is read by an audience of mathematicians, checked for contradictions and usefulness, and only then the proposed system may become widely accepted.   But acceptance of a new language phrase is not checked for contradictions, at least not consciously.   So a self-contradictory phrase like “I could care less” can become popular, even though most of its users do not realize that it is really an ironic use of “I couldn’t care less.”   Language and mathematics are fundamentally different things.

Scientific theories go through a checking process similar to that of mathematics, but there is one major difference -- scientific theories are about the real world, mathematics does not have to apply to any data at all.   It is usually the job of scientists to apply mathematical patterns to observations, only very rarely will a scientist need to invent a new form of mathematics to cover his or her data.   And in that case the usual checking must apply to the new mathematics.  A theory may cover many phenomena and it may utilize more than one mathematical technique.   In this case, the relations that the theory posited among these phenomena also undergoes the usual checking for self-contradictions.   The resulting theory is simply an explanation of the observations.   The purpose of the explanation is to help scientists better understand the observations.   Mathematics aids in this understanding, because it is a set of systems that most scientists agree on and understand.

Of course, careful procedures for accurate data collection must also be observed, otherwise the theories do not apply to the real world.   Data collection is especially tricky in sciences dealing with people, such as linguistics.  You can’t open up people’s heads to find out what’s going on in their brains as they speak, and you can’t deprive children of language by not speaking any human language to them.   But this is no reason to give up on data.   If linguistics is to be a science, it must follow scientific methods.  A linguist once complained that we cannot find all the possible sounds of language by observing all the languages of the world.   What about sounds which have temporarily gone completely out of use?   What about new sounds which may appear someday?  But this linguist was missing the whole point.   Collecting an inventory of all the sounds of language does not explain anything.   However, finding out how often which sounds are used together in various languages and which are incompatible with each other would tell us something about human linguistic capabilities.    Observing only one language would probably not give a large enough sample to say anything about language in general.

Mathematics uses written and sometimes spoken symbols in a way somewhat similar to human language.  Of course mathematical symbols do not cover the enormous spectrum of meaning that words do.  Nor do the mathematical symbols have the psychological impact of words.  Also, mathematical sentences are only statements, they are never questions, commands, etc.

The use of mathematics/logic is usually different from that of language.  Language is used mostly for communication, whereas mathematics is used mostly for thought.  When you speak, you ultimately want to affect another person's mind.  When you do mathematics, you want to keep track of something, for example, counting money,  or checking/proving a mathematical formula.  In other words, you want to affect your own mind.  There is some overlap, of course.  Sometimes we use language to think: if I do this, then she'll do that...  Sometimes we use mathematics to communicate: a scientific paper may compare the speed of a falling stone to the equation V = A x T.  But in general math is for thought, language is for communication.

Linguistics and philosophy often bump up against each other, not always very happily. Over many periods of its history, philosophy has been concerned with the meaning of words, sometimes speculating that all philosophical problems could be solved or at least clarified if only the language being used were clearer. And conversely, many linguists do their semantic research in a way that philosophers have been doing philosophy -- they are concerned with truth-value, intention, sense, and reference.

And some linguists do their linguistics as if they were philosophers -- they feel that linguistics can be done without looking at objective data, but must be done entirely through introspection, that is, asking oneself what is and isn’t grammatical. They claim that each language needs be creatively analyzed into a scientific theory of its grammar.  That you cannot simply observe what is spoken or written or elicit speech from informants and then rely on some general theory to automatically describe a specific language. This means that general linguistic theories covering many or all languages must be raised up one level into meta-theories.  Meta-theories are normally reserved for the philosophy of science, so this creates a crisis as to whether linguistics is really a science or actually a philosophy.  Fortunately, there are objective linguistic data which can be studied statistically, as with other social sciences.  And, given the proper framework of general linguistic theories, this objective data hopefully can be analyzed and "grammars" for individual languages can be created automatically.  Unfortunately, few linguists have had the patience to attempt this.

Physicists can measure gravity, but can linguists measure language? Yes, language can be measured by the amount of information it conveys. This is the subject of information theory. And, even today, information is one of the few aspects of language that can be objectively measured. Information as a quantity is very easy to picture, since we deal with it every day. In general, the longer the sentence, the more information it contains. The longer a word, the higher its information content. And a longer book contains more information than a shorter book. But this is an inexact description of something a little more complicated. The information content of any message is proportional to the amount of symbols that would be needed to efficiently encode it. Usually there are only two kinds of symbols, like the binary code of a computer’s 1’s and 0’s. Information theoreticians use the binary code as a standard of comparison, so they always ask themselves how many "bits" a given message needs to be efficiently encoded.

So how would you encode a sentence into a binary code?  Perhaps you are familiar with binary numbers, so you know that every familiar decimal number can be encoded into a binary number.  Well, you could also assign every letter a binary code, which is what your computer does in order to store and display letters on your screen.  But the ASCII or Unicode used by computers is actually not the most efficient way to store the letters of an English text.  The Morse code (of the telegraph) was more efficient, because it used fewer dots or dashes for the more frequently appearing letters.  For example, an E was only a dot, while an O was three dashes.  Z was dash-dash-dot-dot.  This saved enormous amounts of human energy over the decades when telegrams were the fastest way to convey messages over long distances.   Now, telegraphers used to pause briefly between letters, but such an end-of-letter signal is not necessary if you set up the code correctly.   And this is the way human languages work: although there are no pauses between words, the sound sequence of a short infrequent word will usually not be found within a longer less frequent word.

Communication theory deals with the transmission of information and uses information theory as its basis.  This theory tells us how much information can be transmitted per time through a given channel such as radio waves or an optical fiber cable or sound waves in the air.   It also says that by choosing the right code, you can make communication as reliable as you like by lower the amount of throughput.   This is known as redundancy, and it is an important part of language.   For instance, it has been calculated that each letter in English texts is worth only perhaps a little more than one bit, yet as you have seen above, the Morse code sometimes needed four symbols for each letter.   This means that the most efficient coding for English (taking letter and word combinations into account) would need only about one bit per letter, yet the digital part of the signal alone is perhaps three times more dense than necessary, making communication more reliable.

Humans are not the only forms of life which communicate with each other.   Your own nerve cells fire electronic impulses among themselves, sending various messages throughout your body.   Other specialized cells in other systems also communicate.   Virtually all higher life forms communicate within their species, if only during a mating ritual.   Although single-celled animals and electronic communications devices use signals, only the more advanced species use symbols.  Signals are continuous energy patterns which stimulate a sense organ, whereas symbols are usually discrete items which need to be actively perceived, perhaps by several senses.  Bees use a kind of dancing gesture to inform their sisters where food is to be found.  The dance is partly continuous and partly discrete.  And it is perceived by vibration within the hive, and by sight outside the hive.  Some features of human signs that set our languages apart from most other animal communication are: 
 

Language Feature	Description
Vocal / auditory channel	Speech is transmitted chiefly through sound
Broadcast transmission and directional reception	Speech is spoken by one person and heard by many
Rapid fading	Speech disappears almost instantly after spoken
Interchangeability	All members of species can make the signs
Total feedback	We monitor our own speech output
Specialization	Speech signs are not merely symptoms of some condition
Semanticity	Speech signs do not usually evoke an instinctive behavior
Displacement	Speech signs do not have to be near the things they represent
Arbitrariness	Speech signs do not have to look like the things they represent
Discreteness	Speech signs are not signals, but recognizable entities
Productivity	Speech signs can be combined with other signs into syntagms
Duality	Speech signs make use of the double articulation
Cultural or traditional transmission	We learn speech signs, we do not know them instinctively
Prevarication	We can lie with speech signs
Reflexiveness	Speech signs can refer to themselves
Learnability	We can learn each other’s (foreign) languages.

Animals -- including human animals -- have communication systems because they need to communicate.  If a giraffe has nothing to say to an elephant, then elephants and giraffes are unlikely to share a language.   Humans are social beings, and human language proves it.  Now, it is possible for animals to be fairly intelligent without language.  Without language there is no way for them to share their knowledge and become even smarter.   But without sufficient sociability instinct, animals are not likely to use language, even if they are capable of it.   This may be one reason it is so hard to teach chimpanzees to use sign language -- if they don’t already talk (sign) to each other, why would they want to talk to us?   Currently, there is much debate on whether non-human animals can use language.  Some researchers claim to find various features of human language in animal communication, even when these features are theoretical relics from a time before scientific thinking about language.   Other researchers hold the opposite bias -- they magnify every difference between animal and people communication, apparently in order to maintain the myth that humans are somehow special, at the “top” of the evolutionary mountain.   Perhaps the truth lies somewhere in between -- animals and humans share certain linguistic abilities and don’t share certain others.

The Arts and Culture

Language is also a kind of repository of culture.  Words such as Christmas, luncheon, touchdown, puritanical, wedding, etc., reflect the traditional activities and attitudes of our society.   Presumably, these words would disappear if the cultural concepts behind them were discontinued.   This happened to the word thane,  an extinct semi-noble medieval title.   Then there are expressions and sayings which give us advice on how to behave: A penny saved is a penny earned.   Language records how our society divides up and classifies the items of our world.   A couple of amusing actions words come from the English names for birds: crane and duck.   The Sapir-Whorf hypothesis is certainly true at this level, since it’s almost impossible for any English speaker to go through life never drawing a connection between bobbing your head down and the behavior of certain aquatic fowl.

Sociology

Even within a single language, there can be many levels of speech -- the upper classes may speak one way, the lower classes another, and there may even be a popular middle class dialect.  People striving to move among the classes may change their speech much in the way they would change their dress or manners.    In some countries, written language may differ so much from spoken language as to make them different language.   This was true of Europe in the middle ages, when Latin was the written language, despite the fact that nobody spoke it anymore.   A region with two different language “codes” for two distinct social circumstances is said to be experiencing diglossia.

Remember, too, that language is a social phenomenon used in conversations between individuals.   Speech can be polite, brusque, formal, informal, illogical, humorous, pleasant or intimidating.   You can use any of these variations to control the type of social interaction with your interlocutor or you can become the subject of such control.   Many European languages have two different words for “you”, such as French “tu” and “vous”.   The “T” (intimate) form is used when people are on a first-name basis, the “V” (formal) form everywhere else.   Korean has refined several different levels of formality, plus grammatical honorifics indicating the social status of the person being referred to.   Since these forms are often obligatory, pronouns like “you” and “I” or personal verb endings become unnecessary, and are usually omitted, because people are usually well aware of the social status of everyone they talk about.